All the news from the world of Artificial Intelligence
Perplexity has been on fire recently, and now we have another big announcement from them. Today they unveiled that they are teaming up with Deutsche Telekom to make Perplexity the native AI assistant on their upcoming AI phone.
Nothing quite instils fear in the human heart like the phrase “robot Swarm” Ubtech has unveiled its connected humanoid robotics system that is looking to rival the Helix Network system that Figure AI revealed recently.
It seems that Gemini Live Vision features will be launched on Android later this month for for Gemini Advanced subscribers, the Live Vision feature allows users to show Gemini the environment in real time via the camera.
Anthropic has announced that it has secured an additional $3.5 billion funding from Lightspeed Venture Partners. This sets the company valuation at $61.5 billion. That’s almost 50% more than Elon Musk paid for Twitter in 2022.
After Perplexity announced that they would be launching an agentic browser last week, Opera decided that they would get ahead of the game by announcing their own ‘Browser Operator’. It looks like this will allow users to carry out agent style work with the browser, carrying out tasks on the user’s behalf.
Finally, Grok is coming to Android Android users have been looking at Apple owners with envy up until now as they get all of the cool mobile app goodies and Android gets nothing. But now there is a Grok Android app in beta.
Google is launching a new AI mode experiment in its search tools. The new search mode expands the capabilities of the search itself and the presentation of the information returned. It will start as an experiment that will get rolled out to Google one AI Premium subscribers.
OpenAI is looking to launch a set of new agent models that are designed specifically for professional research tasks. You could use the low level model which is a bargain at $2,000, or if you’re feeling flush, you can get the top tier service for a mere $20,000 per month. This pricing model has been derided by many users who point out how OpenAI have gone from an open model for the good of humanity to a closed model with a $20,000 price tag.
Some users have got early access to Runway’s new Restyle tool. This allows you to upload an original video as the source and then have Runway restyle the video while keeping all of the original proportions and motion.
FreePik has added a fantastic new function to their generative AI tools called Colors. This allows the user to set a particular color palette and have the generated video or image adhere to the colors chosen.
It’s only been a couple of weeks since Convergence launched its AI agent offering called Proxy. They’ve continued to extend their offering since launch and their latest addition is called Template Hub. This is a community library of ready to use workflows that can be shared amongst users.
OpenAI is finally rolling out GPT 4.5 to its plus users. It won’t be available to everyone instantly. It will take some time to roll out, but by the time you’re viewing this, you may already have it.
In robotics we got to see a crazy demonstration of a quadruped learning to skateboard. This comes from a new paper that is exploring a technique called discrete time hybrid automata learning.
X has added a new GROK integration directly into posts. Now you can reply to any post and start it with Grok and you’ll get an AI reply from Grok to whatever you’ve asked. It’s certainly a novel way of integrating GROK use into the timeline, but many users are concerned that it will just flood the replies with Grok slop.
Hedra has launched a new character focused generative video model called Character 3. This model allows you to easily create natural looking animated characters from a single image and and some audio. Many people have been showing off some really cool results with it and it seems to be a big hit.
It seems that the relationship between OpenAI and Microsoft is really turning sour and maybe coming to an end. Today it was revealed that Microsoft are exploring partnerships for an alternative AI provider including xai, Meta, Deepseek and Anthropic. If the relationship is over, the question will be who needed who more.
Sudowrite has launched a new model that is designed specifically for fiction writing called Muse. They claim that all the current AI models suck at fiction writing and they’re going to remove all the typical AI isms from their outputs.
They say their model combines narrative engineering and storytelling intelligence to deliver a higher quality output. Let’s see what storytelling masterpieces it can produce.